0 Up votes0 Down votes

5 views126 pagesAug 22, 2017

© © All Rights Reserved

PDF, TXT or read online from Scribd

© All Rights Reserved

5 views

© All Rights Reserved

- Effect of Pre Straining on the Springback Behavior of the AA5754-0 Alloy
- 05206G_TOC.pdf
- Standart Iso 13314
- Mean Stress Effects in Stress-Life and Strain-Life Fatigue-DowlingBrasilPaper
- Miners Rule and Cumulative Damage Models.docx
- Optimization of the fatigue strength of materials due to shot peening : A Survey
- Din743 Intro English
- TALAT Lecture 2703: Construction and Dynamic Testing of Aluminium Nodes
- tmpCF9A.tmp
- Fatigue
- DOEAL-SG-100_AppM_09-30-05_final
- Corrosion Effects on the Ductile Fracture, Strength and Reliability of Membranes, Plates and Shells.pdf
- FHWA-ICT
- C21_2008_Myers.pdf
- Spring Design_Wayne Book
- Torsion Lab Reviewed (1) (1) (2)
- CC2011_spring_design.pdf
- recos2.pdf
- LTRC Final Report 500 Mechanistic Evaluation of Asphalt Mixtures Containing Thiopave Additives
- 3_Zhang2016

You are on page 1of 126

Theses and dissertations

1-1-2004

uni-axial and multi-axial loading conditions

Sacheen Bekah

Ryerson University

Part of the Mechanical Engineering Commons

Recommended Citation

Bekah, Sacheen, "Fatigue life prediction in a door hinge system under uni-axial and multi-axial loading conditions" (2004). Theses and

dissertations. Paper 55.

This Thesis is brought to you for free and open access by Digital Commons @ Ryerson. It has been accepted for inclusion in Theses and dissertations by

an authorized administrator of Digital Commons @ Ryerson. For more information, please contact bcameron@ryerson.ca.

FATIGUE LIFE PREDICTION IN A

DOOR HINGE SYSTEM UNDER

UNI-AXIAL AND MULTI-AXIAL

LOADING CONDITIONS

by

A thesis

in the Program of

Mechanical Engineering

AUTHORS DECLARATION

ii

ABSTRACT

in the Program of

Mechanical Engineering

Ryerson University

This thesis presents the use of Finite Element (FE) based fatigue analysis to

locate the critical point of crack initiation and predict life in a door hinge system

that is subjected to both uni-axial and multi-axial loading. The results are

design per the standard requirement in the ground vehicle industry. The

preliminary design stage. Numerical analysis also provides the product design

iii

ACKNOWLEDGEMENTS

Dr. Kamran Behdinan and Co-supervisor, Dr. Zouheir Fawaz for their

continuous support and guidance before, during and after the process of this

research. My gratitude also goes to Dr. Kornel Farkas, Jake Xu, Dave McCabe

Cresnik for their priceless support in the Finite Element Modeling and Analysis

aspect of this research and Masoud Alimardani, Simon Chan, Yigui Xu for their

and Van-Rob Stampings Inc. for their joint support in the collaborative research

iv

DEDICATION

Doodave and Woormeela Bekah, my uncle and aunt, Loondeo and Anuradha

Bekah, my brother, Devesh Bekah and my cousins, Nandishi and Dhilan Bekah.

v

TABLE OF CONTENTS

ABSTRACT ....................................................................................III

ACKNOWLEDGEMENTS ..........................................................IV

DEDICATION................................................................................. V

NOMENCLATURE..................................................................... XV

2.1 Overview.................................................................................................... 7

vi

2.3.1 The Stress-Life Curve ...............................................................................................10

APPROACH ........................................................38

3.1 Overview.................................................................................................. 38

vii

3.3 Multi-axial Fatigue .................................................................................. 41

4.1 Overview.................................................................................................. 46

4.5.2 FE Discretizing......................................................................................................... 60

5.1 Overview.................................................................................................. 76

viii

5.3.3 Simulation Results ....................................................................................................84

...............................................................................97

APPENDIX................................................................................... 100

REFERENCES............................................................................. 103

ix

LIST OF FIGURES

Curve.......................................................................................... 16

Figure 2-8 Cyclic Softening and Hardening under Strain Control ............... 19

..................................................................................................... 21

x

Figure 2-16 Graphical Display of Non-proportional Loading....................... 30

Figure 4-3 Maximum Bending Stress for Shell Element (Quad4-node) ..... 51

Figure 4-5 Skew Angle Measurements in Tria and Quad Elements ........... 55

xi

Figure 4-14 Distorted Elements in BS and DS Brackets............................... 63

Figure 4-20 Maximum Principal Stress Contour in the Uni-axial Model ..... 72

Figure 5-10 Loading Histories for the Applied Torque, T1 and T2 .............. 89

Figure 5-13 SWT Curve and The Linear Damage Summation Procedure.... 93

Figure 5-14 Fatigue Life Contour Under Multi-axial Stress State ................ 94

xii

Figure 5-15 Bi-axiality Ratio and Angle of Spread Plots.............................. 96

Figure A-1 Effect of Thickness on the Fatigue Life of the Hinge.............. 100

Figure A-2 Effect of Steels UTS on the Fatigue Life of the Hinge .......... 100

Figure A-3 Effect of Surface Treatment on the Fatigue Life of the Hinge 101

Figure A-5 Effect of Notch Radius on the Fatigue Life of the Hinge ........ 102

xiii

LIST OF TABLES

xiv

NOMENCLATURE

Roman

e engineering strain

E elastic modulus

wedge element

H height of beam

i numerical increment

J jacobian

L length of beam

xv

n number of cycles of operation

S engineering stress

W width of beam

Greek

skew angle

differential change

true strain

t total strain

p angle of spread

true stress

xvi

f true fracture strength

warp angle

xvii

Chapter 1 Introduction

CHAPTER 1 INTRODUCTION

1.1 Objective

The major challenge in todays ground vehicle industry is to overcome

the increasing demands of higher performance, lower weight, and longer life of

components, all this at a reasonable cost and in a short period of time. Fatigue

structural failures [1]. As such, early engineers used many cut-and-try practices

in their machine designs in an attempt to minimize these failures [2]. This has

As the components became more complex, the more difficult it was to apply

structures and obtain solutions faster. These numerical approaches use both the

data obtained from multiple experimental testing together with the analytical

correlations developed.

1

Chapter 1 Introduction

FEA is now an integral part of structural design. FEA allows design analysts

saved.

with locating critical points where cracks might occur in components under

specific loading conditions. This is followed by the estimation of the total life

combine the methods involved in fatigue and FEA to predict the life in a 60-

degree front door hinge assembly system. Finite Element (FE) is used as a tool

to create the virtual model of the hinge and perform a static analysis. The

stresses and strains from the static analysis are then used as input to locate the

critical point and predict the life expectancy of the hinge under uni-axial and

multi-axial loadings. The uni-axial and multi-axial fatigue results are then

only numerical approaches are used to improve the fatigue life of the hinge. As

such, the outcome of these results provide the product development specialists

at Van-Rob Stampings Inc. with tremendous savings in time and cost because of

stage.

2

Chapter 1 Introduction

loads [1]. These fluctuating or cyclic loads induce cyclic stresses in theses

For centuries, it has been known that a piece of metal can fail under

repeated loading at a stress level well below the ultimate tensile strength of the

material. Hence came fatigue, from a French engineer Monsieur Poncelet, who

related fatigue to its biological counterpart meaning tired [2]. However, the first

who conducted repeated loading tests on iron chains [1]. When fatigue failure

Between 1852 and 1870, the German railway engineer August Whler

successfully conducted fatigue tests on full-scale railway axles and small scale

bending, torsion and axial cyclically loaded specimens for different materials

[1-3]. Thereon, he plotted the results of stress amplitude against life cycles to

failure, which is commonly known today as the Stress-Life (S-N) curve. R.R.

Moore later followed the same principles as Whler and modified the S-N

diagram. The currently available S-N data are based under Moores fatigue tests

[1-3].

of fatigue failure rather than just observing the results [1,2]. As a result, in the

late 1950s and early 1960s, two new approaches to fatigue life estimation

3

Chapter 1 Introduction

emerged. The first one is known as the Manson-Coffin Local Strain approach or

predict crack initiation. The second approach is based on Linear Elastic Fracture

Mechanics (LEFM) and attempts to explain crack growth [1]. All the

loading [1,3]. Hence, a new approach emerged known as the critical plane

approach [1-8]. The critical plane approach recognizes that fatigue is essentially

later developed their own theories following the work of Miller [1-3,5-9].

Presently, there is no general agreement about how to fully deal with multi-axial

used in trucks. The 60-degree movement of the hinge embodies the range of

comprising the hinge are a Door-Side (DS) bracket, a Body-Side (BS) bracket

and a circular cross-sectional pin (see figure 1-1) are stamped together. This

allows a free rotation of the DS and BS brackets about the pin, hence simulating

4

Chapter 1 Introduction

the open/close door movement. Each bracket consists of two cut-outs to allow

for fasteners to attach the BS and DS brackets to the body and door of the

vehicle respectively.

II. Computational uni-axial and multi-axial fatigue life predicted for the door

III. The accuracy of the results shows that there is no need for developing costly,

5

Chapter 1 Introduction

chapters four and five are dedicated to numerical analysis, FEA and Fatigue

of the hinge is also portrayed. Chapter six deal with the improvement in the

chapter seven reports the findings and provides a brief discussion and

6

Chapter 2 Fatigue, A Theoretical Approach

APPROACH

2.1 Overview

This chapter focuses on both the basic and advanced forms of fatigue

and traditional theories involved in fatigue behaviour such as the stress-life and

strain-life curves. The onset of uni-axial and multi-axial fatigue behaviours and

uni-axial and multi-axial stress states is provided. The two domains herewith

that the fatigue process embraces two distinct domains whereby the cyclic

stresses and strains follow different paths to failure. In the first domain, the

stresses and strains for the material under investigation are largely confined to

the elastic region. As a result, failure will occur after large number of cycles,

7

Chapter 2 Fatigue, A Theoretical Approach

cycle fatigue. As duty cycles become more severe and components more

Recent investigations have led to believe that high and low cycle fatigue

for the particular material under static loading. The fact is that for an applied

static load, the material will fall in the high cycle region if the magnitude of the

maximum stress is lower than the tensile elastic limit or yield strength. On the

other hand, low cycle fatigue is characterized by a maximum stress higher than

the yield strength. In summary, high cycle fatigue is typically associated with

lives greater than 100,000 cycles whereas low cycle fatigue involves lives

or fluctuating load will fail at a stress level considerably lower than that

nominal stress or S-N approach was the first method developed to describe this

phenomenon. Throughout years, it has been recognized that the S-N method

works best in the high cycle regime where elastic events dominate plastic ones.

8

Chapter 2 Fatigue, A Theoretical Approach

Hence, the S-N curve yields conservative results up to the yield point of the

the low cycle region, where applied strains have significant plastic component.

Bearings

Test Sample

Using a rotating bending test machine, Whler performed tests on notched and

known today as the S-N diagram. Extensive efforts have been made over the

years to understand the fatigue behaviour of metals along the lines of Whler.

More recently, Moore conducted rotating bending tests. Using the R.R. Moore

Fatigue Testing Machine as shown in figure 2-1, fatigue tests were conducted

9

Chapter 2 Fatigue, A Theoretical Approach

Figure 2-2 shows a typical S-N curve based on Moores test. The fact is

that for most materials used nowadays, the S-N curves are characterized by

Moores tests. The usual laboratory procedure for determining the S-N curve is

to test the first specimen at a very high stress, usually about two thirds of its

small number of cycles [1,2]. The stress is subsequently reduced until the point

where no failure occurs, i.e., the endurance or fatigue limit (EL) [1]. It should be

noted that the curve in figure 2-2 does not have the influence of mean stresses,

notches, environment and surface finish. These effects will be discussed later.

The basic equation for the S-N curve as portrayed in figure 2-2 is:

S = SRIi(N f )

bi

(2.1)

where, S is the nominal stress range, SRI is the stress range intercept of the

life line, Nf is the number of cycles to failure, b1 and b2 are the slopes of life

10

Chapter 2 Fatigue, A Theoretical Approach

lines 1 and 2 respectively and i is 1, 2 for the two slopes respectively. The

material is said to have reached the endurance limit (EL). Materials such as mild

material, we have a designated life of 107 cycles as the endurance limit [1].

essentially elastic, meaning that the S-N curve should be used in regions where

lives greater than 100,000 cycles are expected [1]. This is because the S-N curve

is essentially flat in the low cycle region, and would yield an inaccurate

estimation of life. The reason for this apparent flatness is the large plastic strain,

which results from the high load levels. Low cycle fatigue analysis is best

[1,2,4].

important to consider the general types of cyclic stresses that contribute to the

fatigue process. Most fatigue data collected in the laboratory employed fully

a combination of both. Figure 2-3 (b) and 2-3 (c) show a fully tensile loading

11

Chapter 2 Fatigue, A Theoretical Approach

(a)

(b)

(c)

Figure 2-3 Typical Fatigue Stress Cycles1

12

Chapter 2 Fatigue, A Theoretical Approach

stresses has on the fatigue process so that laboratory data can be efficiently

characterize the effect of mean stress on fatigue life have been developed. Of all

the proposed relationships, two has been most widely accepted, those of

According to Goodman, the effect of mean stress will reduce the applied

stress amplitude in a linear way for both tensile and compressive mean stresses

S

S a = S 0 1 m (2.2)

Su

at zero mean stress, Sm is the mean stress and Su is the ultimate tensile stress.

Gerber, on the other hand hypothesizes that the mean stress will reduce

the applied stress amplitude in a quadratic way for both tensile and compressive

state of mean stresses [1,13]. Hence Gerbers relation can be summarized as:

S

2

S a = S 0 1 m (2.3)

S u

Experience has shown that the actual data falls between the Goodman

exists to support one approach over the other. It is therefore recommended that

13

Chapter 2 Fatigue, A Theoretical Approach

Traditionally, the magnitude of the stresses was low, hence the life cycles to

mentioned is strictly confined within the elastic region of the material, before

the yield point. However, as duty cycles have become more severe and the

emerged [1]. In this regime, the cyclic stresses are very high, and a significant

component has short lives in the range of 100 to 100,000 cycles. This type of

fatigue.

locations such as notches, the material response is strain rather than load

controlled [1]. This observation comes from the fact that since most structures

are designed to confine to the elastic regions, critical locations such as notches

the root of a notch will be similar to that of a smooth specimen tested under

similar, their lives will also be similar [1,2,4,5]. A strain transducer attached to

14

Chapter 2 Fatigue, A Theoretical Approach

an hourglass smooth specimen is used to sense and control the appropriate strain

limits. Thus, both stress and strain can be simultaneously monitored throughout

the test, such that the deformation response of a material can be completely

response, to the formation and growth of a critical fatigue crack [2]. Unlike the

curves, namely the cyclic stress-strain curve and the strain-life curve.

monotonically. Tension tests on certain materials provide basic design data such

provides baseline stress-strain curve for evaluating the nature and extent of any

monitored throughout this process until the specimen fails. Figure 2-4 shows a

curve yields the true behaviour of the material when loaded monotonically. In

the latter case, the applied load is divided by the instantaneous cross-sectional

15

Chapter 2 Fatigue, A Theoretical Approach

area. Figure 2-5 shows a true stress-strain curve as compared to its engineering

16

Chapter 2 Fatigue, A Theoretical Approach

point B (see fig. 2-6 below) and the load removed, the curve will follow a

straight line denoted by line BC. The slope of BC is equivalent to the elastic

Line OC represents the zero stress line, positive stress being tensile

be noted that point O is equidistant from points B and D both horizontally and

direction. The curve will then follow the trend DB, forming a complete stress-

strain cycle also known as a hysteresis loop (see figure 2-7). The above

the following four events may happen. Depending on the type of material in

terms of its properties and heat treatment, the material may [1,2,4]:

17

Chapter 2 Fatigue, A Theoretical Approach

cyclically soften

cyclically harden

remain stable

The onset of cyclic softening and hardening is illustrated in figure 2-8 below

where two different materials are tested under fixed strain limits. Although

strain is fixed, the loads vary significantly in that they converge to different

levels.

the cyclically induced strain, thus causing the softening or hardening behaviour.

It is apparent from figure 2-8 that the maximum stress increases with number of

imposed cycles in the case of softening whereby the stress decreases during the

18

Chapter 2 Fatigue, A Theoretical Approach

limit and remain stable until the emergence of a fatigue crack. The hardening

of the ratio of ultimate tensile strength and 0.2% proof strength as follows [1]:

Su

> 1 .4 (2.4)

S 0.2

Su

< 1 .2 (2.5)

S 0.2

For ratios greater than 1.4, the material cyclically hardens whereas a

ratio less than 1.2 demonstrate a softening behaviour. Within the range of 1.2 to

1.4, the material can cyclically soften, harden or exhibit a mixed behaviour [1].

Cyclic Hardening

Limits

Figure 2-8 Cyclic Softening and Hardening under Strain Control1

the materials total life, the hysteresis loop tend to stabilize, meaning the stress

19

Chapter 2 Fatigue, A Theoretical Approach

amplitude remain constant over the entire life portion [1,2,4]. If the stress strain

coordinates relating to the tips of the hysteresis loops are plotted, the locus of

these points generates the cyclic stress-strain curve (see figure 2-9 below) [1].

Unlike the true stress-strain curve, the cyclic stress-strain curve defines

the materials behaviour under cyclic loading conditions. The cyclic stress-

When a material cyclically softens, the cyclic yield strength is lower than the

monotonic yield strength whereas in the case of hardening, the monotonic yield

strength is higher than that of the cyclic loading [1,2]. Figure 2-10 (a) through

20

Chapter 2 Fatigue, A Theoretical Approach

1

n1

t = + 1 (2.6)

E K

where, t is the total strain, is the stress amplitude, E is the elastic modulus,

exponent.

strain curve can be plotted for any material through the above equation.

Figure 2-10 Comparison Between Cyclic and Monotonic Stress-Strain Curves1

21

Chapter 2 Fatigue, A Theoretical Approach

the strain-life (-N) curve. Unlike the S-N curve, the -N curve is composed of

elastic and plastic strain components. If the elastic and plastic strains are plotted

life curve can be generated as shown in figure 2-11 [1,2,4]. Therefore, the

strain-life curve represents the total strain against reversals to failure, total strain

Basquin was the first to derive the mathematical relation between the

Coffin and Manson established the relation between plastic strain and reversals

Manson elastic and plastic components could be combined to form the total

strain amplitude and number of reversals to failure. The strain-life curve based

22

Chapter 2 Fatigue, A Theoretical Approach

1f

t = (2 N )

f

b

+ 1f (2 N f )

c

(2.7)

E

where, the coefficients 1f and 1f are the fatigue strength and ductility

coefficients respectively, 2Nf is the number of half cycles to failure, and the

exponents b and c are the fatigue strength and ductility exponents respectively.

The first and second expressions on the right hand side of the equation are

Like the S-N curve, the -N is equally affected by mean stress. Hence, to

correlate basic fatigue data obtained by testing laboratory specimens under fully

reversed loading with realistic service situations, the -N should be corrected for

mean stress. This can be done by shifting the strain-life curve up or down

2-12.

23

Chapter 2 Fatigue, A Theoretical Approach

The correction for mean stress upon the -N was initially proposed by

Morrow [1,4,15]. By modifying the elastic part of the -N curve, a new relation

for the total strain and number of reversals to failure was developed. Hence, the

( 1

0 )( ) + 1f (2 N f )

f

t = 2N f

b c

(2.8)

E

given cycle [1,4,17]. SWT concluded that the product of the maximum stress

and the strain amplitude would yield a new -N equation including the effect of

mean stress. The maximum stress for a fully reversed loading can be

max = 1f (2 N f )b

(2.9)

f

max a = (2 N )

f

2b

+ 1f 1f (2 N f )

b+c

(2.10)

E

where, max is the maximum mean stress and a is the strain amplitude.

24

Chapter 2 Fatigue, A Theoretical Approach

the SWTs approach has proven to yield more conservative results when the

loading is tensile while Morrows method provides more realistic life estimates

causing the failure is uni-axial in nature. This means that the stress and strain

lie in the plane of the component and the z-axis is normal to that plane. A uni-

The principal and shear stresses normal to the x-y plane are zero [1]. This

means that z = yz = xz = 0.

The aforementioned methods in sections 2.3 and 2.4 are strictly reserved

to uni-axial fatigue analysis. This is because the stress and strain life curves are

created using simple uni-axial testing principles. For the case of the S-N curve,

the Moore testing machine can be used for either axial, rotating or bending.

25

Chapter 2 Fatigue, A Theoretical Approach

loading situations, such that the S-N and -N curves can be usefully employed

to predict life, a new approach was developed. This new approach is commonly

known as the equivalent stress-strain approach. The next major section in this

extent of multiple loading and its impact on fatigue failure before moving to the

equivalent stress-strain approach. Situations where the latter fails are also

briefly discussed.

deformation in at least two of axial, bending or torsion. The induced stress state

varies over time, hence greatly influencing crack initiation and propagation

[1,2,4]. The basic fatigue techniques discussed thus far are based on uni-axial

shafts, connecting links, automotive and aircraft components and many others

involve a multi-axial state of cyclic stress [1]. Adopting the S-N and -N curves

power generation, ground vehicle and aerospace stream have been actively

engaged into research for life predictions under a multi-axial stress state [1]. As

a result, new theories have been recently developed. These theories embrace

26

Chapter 2 Fatigue, A Theoretical Approach

developing stage such that the mathematical correlations available so far have

principal stresses varies with time. To fully comprehend this behaviour through

investigate the two parameters causing it, namely bi-axiality ratio, ae, and angle

Similar to the uni-axial assessment, the stresses and strains at the surface

of the component under investigation are resolved such that the principal and

shear stresses normal to the surface are zero. From the values of x and y , ae

2

ae = (2.11)

1

where, 1 and 2 are ordered with 1 being the most positive of x and y .

27

Chapter 2 Fatigue, A Theoretical Approach

results when ae is zero and p constant. A plot of the principal stresses against

the bi-axiality ratio or angle of spread at every reversal through the loading can

Through a Consideration of Bi-axiality Ratio18

can be seen from the figure that there is some scatter in the plot. The interesting

point to note is that the bi-axiality ratio, ae, tends to align vertically close to zero

[18]. This indicates that the loading is proportional since ae has a constant value

[18]. The large scatter happens at low values of stress and has no significant

impact on the life. Only high values of stress should be taken into consideration.

Likewise, from figure 2-14, the angle of spread, p tend to align itself at about

45 degrees as indicated by the highest spike in the figure. This indicates the

predominant angle with the stress tensor constant at 45 degrees. Again the

smaller spikes should be ignored because they occur at lower stress cycles.

28

Chapter 2 Fatigue, A Theoretical Approach

The onset of uni-axial loading is portrayed in figure 2-15, from which it can be

seen that the stresses line up vertically at a certain value. That approximate

value is significantly close to zero. Hence, the loading is purely uni-axial in this

case.

Through a Consideration of Angle of Spread18

Through a Consideration of Bi-axiality Ratio18

2-16 and 2-17 below. The graphical display in figure 2-16 shows that for each

29

Chapter 2 Fatigue, A Theoretical Approach

a huge scatter in the plot indicating that the principal stress 1 and 2 are non-

proportional to each other at any time during the course of the loading. The

other point of interest can be seen from figure 2-17, which shows how the angle

oscillates between two predominant angles. This indicates that p varies with

Through a Consideration of Bi-axiality Ratio18

There is as yet no general agreement about how to fully deal with non-

developed, but they have not been completely validated [1,18]. Hence, it is

aforementioned methods can be classified as the critical plane approach. For the

current research, the latter was reviewed and briefly explained in section 2.8

30

Chapter 2 Fatigue, A Theoretical Approach

Through a Consideration of Angle of Spread18

state of stress rarely takes place in components rendering the availability of uni-

axial test data useless. As a result, a large number of tests would be required in

which all of the stress components would have to be varied over an entire range

Due the high costs of such tests, it was necessary to develop a theoretical

stress or strain [1]. As a result, the available S-N and -N curves can still be

31

Chapter 2 Fatigue, A Theoretical Approach

used to predict life. The equivalent stresses or strains used are the maximum

principal, the maximum shear and Von Mises [1]. The maximum shear (or

Tresca criterion) and the Von Mises (or Octahedral criterion) stress theories

have gained the widest acceptance [1]. The empirical correlations behind the

aforementioned stress theories are tacitly discussed in the next two sub-sections.

Von Mises predicted that a component would fail under yield when the

second invariant of the stress deviator, J2, exceeds a critical value. For example

if the monotonic yield stress is the failure mode, then the component will fail

J2 =

1

6

[

( 1 2 )2 + ( 2 3 )2 + ( 3 1 )2 ] (2.12)

respectively.

02

J2 = (2.13)

3

Hence the Von Mises prediction of yield in terms of the principal stress

becomes [1]:

[

0 = 0.7071 ( 1 2 )2 + ( 2 3 )2 + ( 3 1 )2 ]

0.5

(2.14)

32

Chapter 2 Fatigue, A Theoretical Approach

On the other hand, the Tresca criterion suggests that a component will

yield when the maximum shear stress under multi-axial loading reaches the

value of the shear stress under uni-axial tension test [1]. The maximum shear

( 1 3 )

max = (2.15)

2

When max exceeds the shear yield stress 0, failure will occur. It is

worth noting that the Tresca criterion is much simpler than that proposed by

Von Mises [1]. However, there is as yet no general agreement which method

prevails over the other. It is all based on the model geometry, number and type

The equivalent stress-strain methods do not take into account the fact

particular planes. In addition to that, the use of multiple loads creates non-

33

Chapter 2 Fatigue, A Theoretical Approach

critical plane approach calculates life by summing damage on each plane at 10-

degree intervals (see figure 2-18). The worst plane is then selected as the critical

available, these methods were not used and will not be further discussed.

The standardized R.R. Moore testing machine and the strain transducer

used to generate the S-N and the -N curves respectively based their tests on

34

Chapter 2 Fatigue, A Theoretical Approach

conditions. Most components are far from being cylindrical and experience

Component size

Type of loading

Effect of notches

The effect of loading type has previously been discussed, whereby the

stress and strain-life curves were corrected for mean stress or strains. Moreover,

the influence of component size can be ignored when finite element is used to

obtain the stresses and strains. However, the effects of notches, surface

the component is a flat plate, there is no difference between nominal and local

stresses. Many realistic components are far from being flat and include holes

and fillets. These holes and fillets can be classified as notches. As a result, there

fatigue life. In order to translate from elastic strains to elastoplastic strains at the

notches, the Neubers rule are applied [1,2,4,19]. Neubers rule asserts that to

35

Chapter 2 Fatigue, A Theoretical Approach

terms of the nominal and local stresses and strains as follows [1,2,4,19]:

K t2 = (2.16)

S e

where, S and e are the nominal stress and strain respectively and and

The nominal stress and strain can be directly measured from the

component itself. In the case of finite element analysis, the local stress and

influencing life [1]. Scratches, pits and machine marks affect fatigue strength

by acting as stress raisers [1]. These stress raisers aid in significant plastic

processes greatly influence the fatigue strength by altering the residual stress at

36

Chapter 2 Fatigue, A Theoretical Approach

compressive layer whereas plating creates a tensile residual stress layer at the

free surface [1]. To properly estimate the life of a surface treated component, it

the current research, the specimen was not surface treated. Hence, the

stresses and strains. To facilitate this process, finite element analysis (FEA) can

geometries are being estimated. It also aids in the estimation of local stresses

and strains. Chapter four is entirely dedicated to FEA, from the modeling aspect

37

Chapter 3 Fatigue, An Experimental Approach

APPROACH

3.1 Overview

prediction. Experimental procedures are widely used nowadays for fatigue life

prediction. However, they have proven to be very costly and time consuming.

chapter, both uni-axial and multi-axial fatigue experiments are conducted prior

FEA and fatigue. This exercise is done in this manner to verify whether FEA

can be applied to fatigue life estimation in the door hinge. This chapter is

respectively.

tests are unavailable or too costly. In the demanding and extremely competitive

many occasions, the time is very limited for product design engineers to meet

customer demands and design an efficient reliable product. In such cases, uni-

38

Chapter 3 Fatigue, An Experimental Approach

investigation, a uni-axial fatigue test is conducted on the hinge such that the

duty life cycle. The uni-axial fatigue test is a hyperextension load against the

full open stop face of the hinge. The experimental set up is shown in Figure 3-1.

This set up consists of the door hinge, the lever bar for load transmission, and a

In the configuration shown in figure 3-1 above, the hinge is in the fully-

39

Chapter 3 Fatigue, An Experimental Approach

opened position, whereby, the BS bracket rests freely on the DS bracket. A load

is induced through the computer, where the former is gradually increased from

zero to a maximum value of 110 Nm. At that maximum value, the load is

instantly removed and the value drops down to zero. As such, the brackets are in

The component is checked for any evidence of crack initiation at every 5,000

cycles.

crack initiation

1 287,000 3.5

2 292,000 5.0

3 295,000 4.5

4 280,000 1.5

5 290,000 1.5

It is seen that the average number of cycles till crack initiates is 289,000.

The crack length shown propagates at a very fast rate from a 1 mm length, in the

crack length. Moreover, the critical location for crack initiation is around the

40

Chapter 3 Fatigue, An Experimental Approach

Figure 3-2 Crack Location Site From Uni-axial Experiment

This section describes the full-scale multi-axial fatigue tests on the door

hinge simulating the actual vehicle environment. However, such tests are very

time consuming. In this case, the door hinge is cycled for an average of 30 days

The multi-axial fatigue set-up consists of the vehicles door (figure 3-3).

The latter is attached to the vehicles body by the upper and lower hinges. The

hinge fixtures, shown in figure 3-4, rigidly attach the doorframe to the vehicle.

The door is driven by a motor that simulates the open/close door cyclic

41

Chapter 3 Fatigue, An Experimental Approach

which the load is induced. The load is applied at 135 Nm for the first 23 cycles

and at 320 Nm on the 24th cycle. The 135 Nm load represents the average torque

other hand, the 320 Nm torque is a hyperextension load when the door swings

Upper Door

Hinge Frame

Door

C.G

Lower

Hinge

Motor

Moreover, at all times during the full cyclical movement, a door weight

environment the door hinge is subjected to in its duty life. The hinge is cycled at

checked everyday.

42

Chapter 3 Fatigue, An Experimental Approach

Hinge Fixtures

table 3-2 below. In the multi-axial case, the average value for crack initiation is

72,000 cycles. Similarly, the crack propagates at a very fast rate. The interesting

point here is the occurrence of crack after only 31,600 cycles in sample four.

location of the crack is at the root of the notch in the BS Bracket (see fig. 3-5).

43

Chapter 3 Fatigue, An Experimental Approach

first evidence of (mm)

cracking (Cycles)

1 88,400 4.0

2 72,000 5.0

3 92,700 9.0

4 31,600 3.0

5 72,000 5.0

BS Bracket

Crack Location

DS Bracket

The average life obtained from the uni-axial fatigue test was 289,000

cycles whereas the multi-axial fatigue test yielded an average life of 72,000

cycles. As previously stated, the life resulting from a uni-axial test is not as

important as the one from a multi-axial test. The latter reflects the actual loading

environment in the door hinge. On the other hand, the former only gives an

44

Chapter 3 Fatigue, An Experimental Approach

the uni-axial fatigue test results are adequate for the product design engineers.

Uni-axial fatigue test results are reliable when the component is still in the

multi-axial tests are imperative. The results clearly showed that the life and

crack location are very dissimilar in each case. At this point, from a validation

point of view, the uni-axial fatigue life should be disregarded. However, for the

current research, the uni-axial result was used as a guideline in the FE modeling

of the hinge. This is because in the early phase of this research, the full-scale

multi-axial fatigue test was unavailable. Hence, the uni-axial fatigue result was

used to refine the gap that existed between numerical and experimental models.

The experimental multi-axial fatigue result was then used to validate that

45

Chapter 4 Numerical Analysis: FE Static

4.1 Overview

In this chapter, Finite Element (FE) is used to model the 60-degree door

hinge. Two models are considered, one for uni-axial loading and the other for

multi-axial loading. The uni-axial model consists of a single hinge whereas the

multi-axial one consists of the entire door hinge system, which includes two

obtain the stresses and strains to be used as input for fatigue life prediction.

certain element types. The result from the aforementioned analysis is then used

as a guideline to select the proper element type for the hinge under

in FE. This is achieved by modeling the uni-axial hinge with two different mesh

options and comparing the effectiveness of both models with the use of

geometry check parameters and the static results. It is observed that the

combination of proper element type and mesh selection would yield static

46

Chapter 4 Numerical Analysis: FE Static

infinitely small decretized domains called elements. They move relative to each

procedure for solving a set of complex equations that describes the physics of

the problem of interest [21,22]. The basic idea behind the FEM is piecewise

entire system into a finite number of small elements, and then approximating the

solution over each element by a simple interpolation function [23]. The primary

solution in a static case is the displacement, u, whereas the stresses and strains

classified as 1D, 2D or 3D. Some of the elements are only applicable to certain

situations and some work better than others in a particular situation. The next

section discusses how elements can be selected based on the simple cantilever

beam example.

47

Chapter 4 Numerical Analysis: FE Static

both are subjected to a bending load). A cantilever beam as shown in figure 4-1

Beam Dimensions

L = 50 mm

W = 5 mm

H = 5 mm

Applied Force

F = 10 N

element. In the latter case, 1D, 2D and 3D elements are used. Table 4-1 below

48

Chapter 4 Numerical Analysis: FE Static

Nodes Node

One-

dimensional

3 (2 beam

1D Bar Linear 2 translations + problems

1 rotation) where

torsional

rigidity is

required.

Three-

dimensional

6 (3 thin-walled

Tria 3 translations + structures.

3 rotations) Should be

avoided due

to high

2D Shell stiffness.

6 (3 Three-

Quad 4 translations + dimensional

3 rotations) thin-walled

structures.

Wedge 6 3 (3 Solid

translations) structures.

3D Solid

Hex 8 3 (3 Solid

translations) structures.

Using the Euler-Bernoulli beam equation, the maximum bending stress can be

calculated as follows:

49

Chapter 4 Numerical Analysis: FE Static

My

max = (4.1)

I

where,

M = FL (4.2)

H

y= (4.3)

2

WH 3

I= (4.4)

12

where, I is the moment of inertia and y is the maximum vertical distance from

respectively. It can be seen from figure 4-2 below that the value obtained for the

maximum bending stress is equal to the one from the Euler-Bernoulli theorem.

Element Edge Length = 10 mm % Error = 0 %

Although ideal for modeling bending situations, the beam element is very

50

Chapter 4 Numerical Analysis: FE Static

limited in the hinge application, such that it is only used in certain structures to

Element Edge Length = 10 mm % Error = 15.8 %

Element Edge Length = 2.5 mm % Error = 3.33 %

Element Edge Length = 1.25 mm % Error = 49.8 %

Figure 4-3 Maximum Bending Stress for Shell Element (Quad4-node)

As a result, shell and solid elements are widely used in place of beam

51

Chapter 4 Numerical Analysis: FE Static

elements. The popularity of the former comes from its six DOF capability.

Since most components are subjected to both forces and moments in realistic

situations, the shell is an ideal choice. Shell elements also reduce the amount of

shell elements fail. One of the most important situations where shell elements

fail is depicted below in figure 4-3. This investigation is based on the same

Figure 4-3 shows the effect of refining the element edge length on the maximum

bending stress.

The effect of altering the element edge length from 10mm to 2.5mm

causes the maximum bending stress to approach the exact solution. Indeed, from

finite element theory, discretizing the structure into smaller elements should

cause the solution to converge to the exact value. However, further refinement

of the mesh from 2.5mm to 1.25mm causes a drastic change in the stress.

bending stress almost doubled from its original value. Furthermore, the contour

plot of the stress distribution as seen from the spectrum is irregular and very

distinct from the previous ones. The occurrence of this result can be explained

length to thickness ratio of 1:2 is appropriate. In this case, the ratio is 1:4 and is

52

Chapter 4 Numerical Analysis: FE Static

because most solids, except for tet4-node elements, do not have rotational

compared to shells. However, as seen in the previous example, shells fail when

Element Edge Length = 10 mm % Error = 12.5 %

Element Edge Length = 2.5 mm % Error = 0.83 %

Element Edge Length = 1.25 mm % Error = 2.92 %

53

Chapter 4 Numerical Analysis: FE Static

beam) cannot be modeled using shell elements. The use of solids is unavoidable

in those situations. The next example shows the cantilever beam modeled as a

solid or commonly known as the brick element. The results show that using a

refined mesh through the length and thickness cause the solution to converge to

the exact value. The interesting point to note here is how the unavailability of

The results based on this exercise are used to model the hinge system.

Before discussing the detailed finite element modeling and analysis of the hinge

improper elements. The next section discusses, and provides acceptable data

to its skew and warp angles, together with aspect, taper and jacobian ratios. All

these measures represent the amount that the solid element deviates from its

ideal shape. In practice, highly skewed and warped elements, as elements with a

referring to each face in the element. More precisely, the skew angle is

measured on each face as if it is a tria or quad element. Figure 4-5 below shows

the measurement of skew angle for a tria and quad element respectively. Hence,

54

Chapter 4 Numerical Analysis: FE Static

the skew angle for a brick or wedge is taken as the highest angle resulting from

the respective quad or tria element. In practice, this angle should be smaller than

30 degrees [22,23].

Figure 4-5 Skew Angle Measurements in Tria and Quad Elements

referring to each face in the element, whereby the highest angle is reported as

the warp angle. Figure 4-6 shows a highly warped quad element with as the

warp angle. The warp angle should be smaller than 5 degrees [22].

opposing faces. Unlike the skew and warp angle, the aspect ratio is calculated

differently for a brick and a wedge elements. For a wedge element, the two

triangular faces are averaged to obtain a mid-surface. The aspect ratio of the

55

Chapter 4 Numerical Analysis: FE Static

latter is then calculated. For the remaining quad faces in the wedge element, the

ratio of the maximum to minimum length are measured and multiplied by the

aspect ratio of the mid-surface tria element [22]. The resulting solution is the

aspect ratio of the wedge element (see figure 4-7 below). The maximum

3 h2 h4 3 h2

Tria Aspect Ratio = Wedge Aspect Ratio =

2 h1 h3 2 h1

The Aspect Ratio (AR) in a brick element is calculated as the ratio of the

distance between opposing faces. The distances between the centerpoints of all

three pairs of opposing faces are compared and the maximum value taken as the

The taper and jacobian ratios only exist for quad or brick elements. The

former is calculated by splitting the quad element into four triangles connected

at the mid-point of the quad. The area of each triangle is calculated. The ratio of

the smallest to the total area is the taper ratio. Therefore, for the brick element,

the aforementioned procedure is repeated for each face and the largest value is

reported as the taper ratio of the brick element (see figure 4-9 (a)). A minimum

56

Chapter 4 Numerical Analysis: FE Static

max(h1, h 2, h3)

Brick AR =

min(h1, h2, h3)

of Face 2

4 min(a1, a 2, a3, a 4) J2

Taper Ratio =

a1 + a 2 + a3 + a 4 Jacobian Ratio =

J1

(a) (b)

Figure 4-9 Taper and Jacobian Ratio Measurements in Brick Elements

The jacobian ratio is a measure of the deviation of a quad element from its

ideal rectangular shape. The jacobian is calculated for each face in the brick

element. The jacobian ratio is the maximum ratio of jacobian between the

opposite faces. An ideal element has a jacobian ratio of 1.0. Generally, jacobian

ratios of 0.7 and above are acceptable (see figure 4-9 (b)) [22].

57

Chapter 4 Numerical Analysis: FE Static

assembling. Two distinct models are considered. The first model (model A) is

elements and the second (model B), a bi-parametric solid, consists solely of

fatigue life prediction, is compared based on the FE static results. Two separate

static analyses are then performed on each model. The first one is to compute

the stresses and strains and use them as input data for a uni-axial fatigue

analysis whereas the other is for a multi-axial fatigue analysis. The uni-axial and

multi-axial models are quite dissimilar from each other in the sense that the

former consists of a single hinge whereas the latter, the entire door hinge

The hinge model, created in the Computer Aided Design (CAD) package

format contains the solid geometry of the hinge (see figure 4-10 below).

bracket and Pin are symmetrical components unlike the DS bracket that has an

oval and a circular shaped hole on each side of the component. The Pin is

removed from the model since it does not contribute directly to the FEA. The

pin mechanism allows rotation of the DS about the BS bracket, thus transferring

58

Chapter 4 Numerical Analysis: FE Static

loads from one bracket to the other. A different approach, as will be seen later in

Front View

Bottom View

The mid-surface being equidistant from the top and bottom surfaces of the CAD

model. Model B is split into bi-parametric surfaces for reasons that will become

59

Chapter 4 Numerical Analysis: FE Static

4.5.2 FE Discretizing

excessive stiffness. Rectangular elements (Quad4) are ideal for shell mesh.

difficult to avoid Tria3 elements. It should be noted that a model in which the

total number of Tria3 is less than five percent of the total number of elements in

the model yields conservative results [22]. Figure 4-11 shows the discretized

shell model with Quad4 and Tria3 elements with an element edge length of 3

mm. It can be seen from figure 4-12 that Tria3 are indeed formed at curvature

and holes.

The shell elements are then extruded to create solid elements. This

60

Chapter 4 Numerical Analysis: FE Static

model in each direction normal to the surface of the shell element. As a result,

Tria3 and Quad4 become Wedge6 and Hex8 solid elements respectively. It

should be noted that the original surfaces couldnt be meshed directly by solid

are used, gaps are induced in the FE model. This is because at curvatures, the

original surfaces are to be used for meshing, extruding a shell mesh (model A)

created.

61

Chapter 4 Numerical Analysis: FE Static

isomesh with hex8 elements. From section 4-3, it was seen that to take into

account bending, the solid elements had to be refined through the thickness. At

least 3 solid elements should be used through the thickness, if proper bending

effects are to be included [22]. Figure 4-13 shows a comparison between the

are performed to verify whether the elements qualities are acceptable. Figure 4-

62

Chapter 4 Numerical Analysis: FE Static

14 below shows the potential areas where the elements failed the skew angle

error messages are also issued showing the extent by which the element failed.

Referring to section 4-4, where the minimum or maximum acceptable values for

each parameter are listed, the extent by which the failed element deviated from

the acceptable values can be calculated. Table 4-2 below displays the outcome

63

Chapter 4 Numerical Analysis: FE Static

Bracket Total % % % % %

type and number Failure Failure Failure Failure Failure

model of in in aspect in taper in skew in

elements jacobian ratio test ratio test angle warping

ratio test test test

Worst: Worst: Worst:

0.748 34.30 15.00

Worst: Worst: Worst:

0.753 36.70 15.00

Worst: Worst: Worst:

0.130 83.30 15.00

Worst: Worst: Worst:

0.693 77.20 15.00

Although not all the elements in the models satisfied the geometry check

criteria, when meshing the structure, the goal is to minimize the number of

distorted elements and to obtain values for these distorted elements that are

close to the acceptable value. It can be seen from table 4-2 that all the elements

pass the jacobian and aspect ratios. Moreover, the percentage failure in taper

ratio arising from models A and B are very close. However, the value for the

worst element in model A is very close to the acceptable value. On the other

64

Chapter 4 Numerical Analysis: FE Static

hand, in model B, the taper ratio for worst element deviate quite a lot. The skew

angle check can be analyzed in a similar fashion. It can be seen that the

elements in model B are highly skewed with skew angles significantly larger

models because of the highly curved surfaces. The percentage of warp angle

The objective behind the geometry tests in the two models is to compare

seen that model A is more accurate. The accuracy of the FE model can also be

certain point in the model other than notches and loading areas arise from

distorted elements. This can result in the prediction of wrong stress values. This

The final step involves assembling all the discretized components of the

hinge. To this end, the two components of the hinge, the BS and DS brackets,

are joined using multi-point constraints (MPC). MPC acts as a rigid bar between

two nodes and constrains all DOF, three translational in his case, so that the

nodes move by the same amount in every direction. One of the two nodes is

called a dependent node and the other, an independent node. If the independent

similar displacement. Therefore, MPC can be used to transfer loads from one

node to the other. In this case, the MPC is used to replace the Pin connecting the

65

Chapter 4 Numerical Analysis: FE Static

versa. Since MPC is rigid, it will not deform under external load and does not

The interesting point to note here is the mechanism behind the motion of

the brackets in the hinge. In reality, the DS bracket rotates about the Pin with

degrees. This motion simulates the open/close door movement. Hence, at the

zero-degree position, the door is fully-opened and at the 60-degree position, the

door is closed. Figure 4-15 shows the range of motion of the hinge.

The mechanism of the hinge motion from the fully-closed to the fully-

66

Chapter 4 Numerical Analysis: FE Static

opened position is very complex to model through FE. This is because contact

MSC.MARC was used to model the contact points. Two types of contacts were

used, rigid and sliding contacts. This exercise proved to be very tedious, time

If the rigid body motion of the hinge is considered, the stress is negligible

A rigid contact can then be used to constrain the BS and DS brackets when

Rigid MPC

constraining the

clearance between the

BS and DS brackets

Assembled Component

Rigid MPC replacing

the Pin

Based on these assumptions, MPC were used to model the rigid contact.

These rigid MPC ties the nodes on the BS and DS brackets together at the

67

Chapter 4 Numerical Analysis: FE Static

bracket through the MPC. Figure 4-16 displays the use of MPC to constrain the

hinge.

conditions. Therefore, two separate models are built for the different loading

conditions. In the uni-axial model, the circular and oval holes in the DS brackets

are fixed and a force is applied via a block on top of the BS bracket. Although

MPC could be used to apply the load, a meshed block is used instead to reflect a

more realistic loading. The block is attached to the BS bracket using rigid MPC.

All DOF Fixed Applied Load = 335 Nm

Applied Load = 110 Nm Door Weight = 480 N

Uni-axial Model Multi-axial Model

The multi-axial model consists of two hinges and a T-shaped support to reflect

the entire door-hinge system. In this model, the upper circular holes on the BS

68

Chapter 4 Numerical Analysis: FE Static

brackets are fixed. Two separate load cases are applied to the multi-axial model,

namely an applied torque and the weight of the door. The applied torque is

converted to a force because solid elements do not have rotational DOF. The

uni-axial and multi-axial models are shown in figure 4-17. When constraining

the holes in both the uni-axial and multi-axial models, a certain area around the

holes was fixed instead of the circumference only. Hence, the correct bolt and

fatigue life in the door hinge subjected to uni-axial and multi-axial stress states.

and use these as input to calculate fatigue life. Before implementing a static

elements in the hinge. The geometric properties in this case are self-defined

when solid elements with a constant gauge are used (E.g., if shell elements are

property). On the other hand, material properties have to be defined in all cases.

In this case, the material used is steel (SAE Grade 1008-1010). The desired

static results from Finite Element Analysis (FEA) are used as input data to

properties of the material, since it is unknown at this point whether the stresses

and strains will fall in the high or low cycle region. The cyclic and monotonic-

69

Chapter 4 Numerical Analysis: FE Static

The question here is whether to use the monotonic or cyclic curve. From a

static analysis point of view only, the monotonic curve should be used.

However, since the stress and strain values are used as input data to calculate

fatigue life, the cyclic curve is used. This feature is explained in chapter 5.

MSC.NASTRAN is used to solve for the stresses and strains. This analysis is

conducted on both models A and B. The results are discussed in the next

section.

The question here is whether to use the monotonic or cyclic curve. From a

static analysis point of view only, the monotonic curve should be used.

However, since the stress and strain values are used as input data to calculate

fatigue life, the cyclic curve is used. This feature is thoroughly explained in

conducted. MSC.NASTRAN is used to solve for the stresses and strains. This

70

Chapter 4 Numerical Analysis: FE Static

analysis is conducted on both models A and B. The results are portrayed in the

next section.

In this section, the stress contour plots for both the uni-axial and multi-

axial models are displayed. FEA generates maximum and minimum principal

stresses (i.e., tensile and compressive stresses) together with the equivalent Von

Mises and shear stresses. It is important at this point to analyse each stresses

effectively and use the most critical one as input to calculate life. The rule of

thumb is that the critical stress will have the most adverse effect on fatigue life.

Using improper stress values will result in inaccurate fatigue life prediction.

investigated. The most appropriate model can be selected (see figure 4-19) by

In a uni-axial stress state, the X- and Y- components of the stresses lie in the

Model A Model B

Figure 4-19 A Comparison between the Z-Component of Stress

In Models A and B based on a Uni-axial Loading

71

Chapter 4 Numerical Analysis: FE Static

It can be seen from figure 4-19 that the stress is uni-axial in nature in

both models except at the holes on the DS brackets. The sharp transition in

stresses in these areas is due the constraints. The hinge was fixed in all DOF at

these locations, thus the high stress values are really artificial stresses. In an

experimental analysis, the stresses will be close to zero at these locations. One

of the drawbacks of FEA is that artificial stresses are created at loading and

spectrum on the right hand side of both models, it can be seen that the stresses

72

Chapter 4 Numerical Analysis: FE Static

for uni-axial and multi-axial fatigue analysis. Figures 4-20 and 4-21 display the

maximum principal stress based on the uni-axial and multi-axial loading cases

respectively. In the uni-axial model, the maximum stress is 255 MPa, and in the

multi-axial model, the maximum stresses are 343 MPa and 142 MPa for the

torque and weight cases respectively. The maximum principal stress is the most

has an equal element edge length and thickness of 2.5 mm. A mesh size of 3.5 is

initially used, and subsequently reduced to 3 mm and 2.5 mm. Although, the

results are not reported here, the stress results tend to converge. However, the

could not be further refined. This is because the high mesh density required a

stress values were taken as the converged results. At this point, it is assumed

that a correct life prediction by fatigue analysis implies that converged stress

values are employed. The most important observation from the static analysis is

the location of the maximum stress in the model for both the uni-axial and

multi-axial models.

73

Chapter 4 Numerical Analysis: FE Static

Figure 4-21 Maximum Principal Stress Contour in the Multi-axial Model

74

Chapter 4 Numerical Analysis: FE Static

Firstly, in the uni-axial model, the maximum stress occurred at the inner

edge of the oval-shaped hole. Although this is the constrained area, only the

stresses at the top and bottom surfaces can be ignored (i.e. surfaces in contact

with the washer). It will be seen later that crack is initiated at the oval hole. At

this stage, no further explanation will be given until confirmation from the

model are at different locations for the torque and weight cases respectively. It

will be seen afterwards that the stresses in both cases are superimposed to

simulate the multi-axial environment of the hinge. Similarly, it will be seen that

Indeed, this is the real advantage of FEA. It is not only efficient in the

at the root of notches if a sufficient small size mesh is used. The maximum

principal stresses are therefore used as input data in the numerical fatigue

analysis.

75

Chapter 5 FE Based Fatigue Analysis

5.1 Overview

In this chapter, the fatigue life of the door hinge is predicted using

former, the hinge is subjected to a single load and the stress results associated

with this load are cycled until crack initiates. In the latter, a hinge system is

analyzed, whereby results from two separate loads are superimposed and used

to predict life. Moreover, the bi-axiality ratio and angle of spread are evaluated

and plotted at the critical location in the multi-axial case to assess whether a

The onset of high and low cycle fatigue are explained in chapter two. It

is stated that if the maximum stress in a component is lower than the yield

strength of the material, the component fails under high cycle fatigue. This

means that the life cycles to failure is greater than 100,000 cycles. In contrast,

when the maximum stress is higher than the yield strength, the component fails

under low cycle fatigue with life between 100 to 100,000 cycles. As a matter of

fact, knowing the failure mechanism before running the fatigue analysis is very

important. This is because in the high cycle region, using parameters based

upon the S-N curve yields conservative results. Similarly, in the low cycle

76

Chapter 5 FE Based Fatigue Analysis

Consider the maximum stress in the hinge in the uni-axial loading case

and the S-N curve for the material used, SAE1008 hot-rolled (see figure 5-1).

The maximum stress obtained from FEA is 255 MPa while the stress range on

the vertical axis of the S-N curve is varying from 80 MPa to 256 MPa. Since the

maximum principal stress is right at the yield point and lies within the stress

range, using the S-N curve would result in the prediction of conservative fatigue

life (line LA1 in figure 5-1). On the other hand, consider the maximum principal

strain and the -N curve (see figure 5-2). The corresponding maximum principal

strain is 3.63E-3. If this strain value is applied to the -N curve (i.e., using the

Strain-Life approach), the number of life cycles to failure will be infinite (line

LB1 in figure 5-2). Hence, the S-N approach is the most appropriate in the uni-

Likewise, consider the maximum principal stress and strain in the multi-

axial loading case. Since the torque yields the highest stress and strain, it is

maximum principal stress and strain are 343 MPa and 8.5E-3, respectively. If a

horizontal line is drawn from the stress value of 343 MPa to the far right vertical

axis on the S-N curve, it is seen that the line does not cross the curve (line LA2

in figure 5-1). Therefore, an infinite life will be predicted if the S-N curve is

used. On the other hand, if the corresponding strain value is used on the -N

curve and a similar line drawn from the maximum principal strain, the line

crosses the curve at a finite value (line LB2 in figure 5-2). This shows that the

77

Chapter 5 FE Based Fatigue Analysis

fatigue analyses separately, since they require different analysis parameters. The

next two sections describe the uni-axial and multi-axial fatigue analyses,

78

Chapter 5 FE Based Fatigue Analysis

respectively.

chart below (figure 5-3) shows the required path for a successful prediction of

life. Based on the observations in the previous section, the S-N approach is used

for the uni-axial fatigue analysis. The available S-N curve is characterized from

loading. The hinge differs significantly from the test specimen in geometry,

impact on fatigue life. Hence, the S-N curve is modified accordingly so that it

experimental process. This means that the static load applied to the hinge is

The loading history for the hinge in the uni-axial fatigue analysis is

obtained experimentally. Figure 5-4 shows the loading history. In general, the

range of any loading history is from 1 to +1. For a fully tensile loading, which

is the case here, the minimum value should be greater than zero and the

maximum value no greater than +1. In the uni-axial experiment, the applied

79

Chapter 5 FE Based Fatigue Analysis

RESULTS HISTORY INFORMATION

1 2

scaled according to the compared to the fully reversed

maximum stress. sinusoidal curve.

The maximum value of If the mean stress is non-zero, the

the scaled loading history S-N curve is modified to take into

is the maximum principal account mean stress.

stress Goodman or Gerber mean stress

correction is applied to the S-N

curve.

by the rainflow cycle process [1,2,25-27].

The rainflow process converts the load

history into blocks of constant stress

amplitude [1,2,25-27].

The modified S-N curve is used to calculate

the damage caused by each block of stress.

The damage is accumulated based on

Miners rule (i.e. the life is calculated at each

cycle) [1,2,4,28].

When the sum of the damage equals unity,

the accumulation stops.

The life at the unit value is taken to be the

fatigue life.

LIFE

80

Chapter 5 FE Based Fatigue Analysis

Hence, the true loading history has a minimum and a maximum value of

loading history is normalized (i.e. the true loading history is divided by the

maximum value of 110 Nm). The static stress resulting from the applied load of

110 Nm is then superimposed with the normalized loading history. As such, the

The next step in the analysis is the implementation of the S-N curve. As

previously mentioned, the latter has to be modified in case mean stresses are

81

Chapter 5 FE Based Fatigue Analysis

clearly seen that mean stresses are present. Therefore, based on either Goodman

or Gerber mean stress correction factors, the S-N curve is modified. This can be

done by pivoting the elastic line about the intersection point on the horizontal

the other. Therefore, both Goodman and Gerber corrections are used and the

one yielding the most conservative result is taken as the appropriate correction

factor. The final step involves combining the scaled loading history and the

stress amplitudes such that it can be used on the S-N curve. This conversion is

82

Chapter 5 FE Based Fatigue Analysis

After the rainflow histogram is created, the life is calculated via the

linear damage summation based upon Miners rule [1,2,4,28]. Miner calculated

the life by summing the damage after each cycle. For each block in the

histogram, the corresponding life is extracted from the S-N curve and a damage

the number of cycles of operation to the total number of cycles to cause failure

at that stress level [1]. This can be mathematically expressed as follows [1]:

83

Chapter 5 FE Based Fatigue Analysis

n1

D1 = (5.1)

N1

where, n1 is the first cycle and N1 , the corresponding life cycles to failure

obtained from the S-N curve from S1. Hence, failure is predicted when the sum

D

i =1

i =1 (5.2)

damage parameter at which the accumulation stops (i.e., if the damage sum

D1000).

stress/strain combination, material curve and mean stress correction. The uni-

superimposed with the loading history and the latter is cycle-counted by the

called the rainflow histogram. Moreover, the loading history is compared with a

sinusoidal waveform to assess the presence of mean stresses in the hinge. The

corrected S-N and the rainflow histogram are simultaneously used to calculate

84

Chapter 5 FE Based Fatigue Analysis

damage using Miners rule. The fatigue life cycles to failure is then evaluated

from the accumulation of damage. Figure 5-8 below shows the fatigue life

contour in the hinge. The minimum life cycles to failure is shown in a log scale

The weakest spot is around the edge of the oval hole in the DS bracket.

The minimum life cycles to failure is 309,000 cycles. It should be noted that if

the analysis is conducted as described, the minimum life is 383,000 cycles. This

reality, the edge of the hinge has a rough surface due to manufacturing

hinge and the average value implemented in the analysis. That average value is

As previously mentioned, the location of the hot spot is around the oval

hole of the DS bracket. More precisely, crack occurs at the contact surface

between the bolt and the edge of the bracket. This can be explained by

considering the sliding motion of the bolt around the oval hole.

In contrast to the circular hole, the bolt does not fit perfectly

around the edge of the oval hole. The clearance is to allow easy placement and

removal of the hinge from the vehicle. Thus, when the load is applied and

cycled, the bolt slides across the oval hole. Although the sliding movement is

scale, which eventually turn into crack. Such mechanisms are carefully modeled

85

Chapter 5 FE Based Fatigue Analysis

in the FEA. The maximum stress located on the top or bottom surface of the

hole (i.e., areas in contact with the washer) were ignored in the analysis.

Figure 5-8 The Uni-axial Fatigue Life Contour in the Hinge

86

Chapter 5 FE Based Fatigue Analysis

The life predicted from the numerical analysis was 309,000 cycles

numerical point of view, fatigue is usually predicted in the order of two or three

times the realistic value. In this case, life was predicted accurately. The required

fatigue life is 260,000 cycles was reached. However, the environment in which

the hinge operates is far from being strictly uni-axial. The service environment

conducted.

used. This assumes a uni-axial stress state when calculating life. The bi-axiality

ratio and angle of spread is evaluated at critical locations in the hinge system in

loading exist. For proportional loading, the uni-axial assumption is valid and the

has to be re-evaluated by the critical plane approach. Figure 5-9 shows a flow

Moreover, from static FE results, it is seen that the stress and strain

values fall in the low cycle region. Hence, the equivalent stress-strain approach

uses the - and -N curves to evaluate life. In the next section, it is seen how

the multiple load cases and their respective loading histories are combined.

87

Chapter 5 FE Based Fatigue Analysis

LOAD CASE 1, 2, .N

1 2

FE STATIC MATERIAL LOADING

RESULTS INFORMATION HISTORY

SUPERPOSITION

OF LOAD

HISTORIES

The stress vs. strain at each 3

node is plotted and compared

with the baseline - curve

obtained from smooth specimen The un-scaled loading history is

testing. compared to the fully reversed

Based upon this comparison, sinusoidal curve.

Neubers rule is asserted to If the mean strain is non-zero,

correct for plasticity at the the -N curve is modified again

notches. to take into account mean

As a result, the baseline - and strain.

-N curves are modified SWT or Morrow mean strain

according to the most critical correction is applied to the new

load case. -N curve.

5 by the rainflow cycle process.

The rainflow process converts the load

Assess Multi-axiality history into blocks of constant strain

amplitude.

The modified -N curve is used to calculate

the damage caused by each block of strain.

The damage is accumulated based on

6 Miners rule (i.e. the life is calculated at each

cycle).

When the sum of the damage equals unity,

the accumulation stops.

The life at the unit value is taken to be the

LIFE fatigue life.

88

Chapter 5 FE Based Fatigue Analysis

is subjected to in its duty life cycle. The loading history for the hinge system is

the open/close door movement. This means that the loading history starts at the

However, since the full range of motion of the door is not considered, the

loading history was modified. The loading histories shown in figure 5-10 are the

outputs from the actual testing process. The loading histories in the multi-axial

process, the induced loading consists of two torques. The first torque (T1) is at

135 Nm and the other (T2) at 335 Nm. As such, T1 is cycled 23 times before

P2 P2

P1 P1

Henceforth, the unit of life cycles to failure is entered as 24 cycles. Hence, each

89

Chapter 5 FE Based Fatigue Analysis

the closed position of the door whereas the maximum point on each curve

The combined loading history has close to a million data points. As seen

in figure 5-11, each loading torque has small spikes. Those spikes are the result

of vibration in the hinge when the door swings from a closed to an opened

position. If the damage is summed at each data point, the simulation time will

be very lengthy. To reduce the simulation time, a process called peak valley

slicing is used [1,2]. Peak Valley Slicing (PVS) is a fairly simple mechanism,

which tracks and extracts the peaks and valleys of all signals to be used in the

analysis. Figure 5-11 shows the process of PVS, whereby the original complete

loading consisting of one million data points is converted to one with only 292

points. When a gate value is entered, which is usually a small percentage of the

load, the peaks and valleys that fall within that gate value are deleted. Hence,

the torques, the weight does not change throughout the open/close door

Similar to the S-N approach in the uni-axial fatigue analysis, the load

90

Chapter 5 FE Based Fatigue Analysis

history is compared to the sinusoidal waveform and corrected for mean strain.

Prior to that, the loading histories are combined with their respective static load

case strain results. Furthermore, the two load histories from the combined

torques and weight are superimposed and resolved to a single scalar loading

history.

Step 1

Step 2

91

Chapter 5 FE Based Fatigue Analysis

the use of the - and -N curves. In addition, Neubers rule is used in the low

cycle region to correct for plasticity by shifting both - and -N curves upward.

[1,2,4,19]. The amounts the curves are shifted depend on the extent of plasticity

involved. Henceforth, the stress and strain results from FE are used to determine

the amount of plastic deformation in the hinge. The stress vs. strain result is

plotted at each node from the FEM and the resulting plot compared with the

baseline - curve obtained from the smooth specimen testing. This is the main

reason behind the use of cyclic stress-strain in the FE static analysis. Hence, the

stress vs. strain plot has the shape of the predefined - curve and can be

accurately compared with the corresponding baseline curve (see figure 5-12).

- Curve -N Curve

1

2 4

2- Original - Curve 4- Original -N Curve

process is similar to the uni-axial S-N approach, where the loading history is

compared to the sinusoidal waveform. Since the loading history is tensile, the

92

Chapter 5 FE Based Fatigue Analysis

SWT correction factor is used. SWT corrects for mean strain by pivoting the -

N curve about the zero-life vertical axis (see figure 5-12). Miners rule is then

used to sum the damage and predict life (see figure 5-13).

Figure 5-13 SWT Curve and The Linear Damage Summation Procedure

In summary, the maximum principal strain for each load case (applied

torque and weight) is combined with the respective loading histories. The

constant strain and the modified -N are used to predict life. A multi-axial

shows the fatigue life contour in the hinge system. It is seen that the crack

location is on the bottom surface of the DS bracket with a life of 10,470 cycles

to failure. The log of life is shown on the bottom right corner of figure 5-14.

contact with the washer. The location of crack at that location is spurious since a

93

Chapter 5 FE Based Fatigue Analysis

surface in contact with a washer cracks only in the presence of machine marks

or extrusions. This is not the case here, meaning the location and estimation of

life is incorrect. The occurrence of a crack at that location is due to the presence

of artificial stresses from FE static analysis. The contact surface of the washer

and the doorframe was constrained with rigid MPC. As a result, the MPC acts

as links that concentrate loading at specific nodes, thus causing the high

stresses. The stresses and strains at loading and constrained nodes should be

94

Chapter 5 FE Based Fatigue Analysis

The real hot spot is around the flange in the BS bracket. More precisely,

the crack location is at the notch radius. The minimum life cycles to failure at

that location is 99, 600 cycles. The initial value was calculated as 119,000

bi-axiality ratio and angle of spread at the critical location. Figure 5-15 (a)

displays the plot of Maximum principal strain vs. bi-axiality ratio at every

reversal in the loading history. Similarly, figure 5-15 (b) shows a plot of

It can be seen from figure 5-15 (a) that the bi-axiality ratio tends to line

up vertically, not equal but close to zero at the most critical node. This indicates

by examining at the angle of spread (figure 5-15 (b)), which shows that the

experimental value of 72,000 cycles. Moreover, it can be seen that the accepted

standard value of 260,000 cycles was not reached. This means that the hinge has

sensitivity analysis in the hinge to improve the life. The sensitivity analysis was

95

Chapter 5 FE Based Fatigue Analysis

Bi-axiality Ratio

Reversal

Angle (Degrees)

96

Chapter 6 Conclusion and Recommendations

RECOMMENDATIONS

6.1 Conclusion

FE based fatigue was used to locate the critical point of crack initiation

and to predict the life in a door hinge system. Both uni-axial and multi-axial

models were built, for uni-axial and multi-axial loading cases, respectively.

Also, two separate FE analyses were conducted in each loading case; static and

fatigue. The static analyses were conducted prior to fatigue and the stress and

strain results were then used as input to fatigue life prediction. The simulation

results for the uni-axial loading case yielded a life of 309,000 cycles compared

was obtained compared to the experimental value of 72,000 cycles for the multi-

The accuracy of the simulation results showed that the FE based fatigue

life prediction approach was very effective. Hence, this approach can be

the preliminary design stage. The simulation results also provide the product

The standard requirement for the durability of such a hinge in the ground

97

Chapter 6 Conclusion and Recommendations

vehicle industry is 260,000 cycles. From the multi-axial results, it is seen that

the life of the hinge was distinctly lower than the standard requirement. Hence,

the hinge. However, due to the strong influence of cost in design optimization, it

was seen that the life could only be altered by modifying the notch radius of the

hinge. The life was only increased to 189,000 cycles under the multi-axial

loading case. Although the required design life was not attained, the

improvement in fatigue life enables the replacement of the door hinge only once

life up to a factor of two. Accurate results are obtained in very few cases, as is

for the door hinge. The statistical nature of fatigue is illustrated in chapter three,

where it is seen that the lives evaluated experimentally varies from 72,000 to

92,700 cycles for the multi-axial loading case. The cause for this discrepancy is

mainly the surface roughness in the hinge. During the manufacturing process,

each hinge has a different surface roughness associated with it. Therefore, for

future research, more hinge samples should be tested and their surface

be found between the experimental and simulation results (see figure 6-1). This

98

Chapter 6 Conclusion and Recommendations

correlation can then be used to evaluate fatigue life in different hinge designs or

1.2

and Numerical Results

1

Life (10E5)

0.8

0.6

0.4

0.2

0

0 2 4 6 8 10 12

Surface Roughness (10E-6 inches)

Experimental Numerical

improve the life of the hinge. Hereafter, a sensitivity analysis was conducted on

the hinge. The analysis included the effects of thickness, materials UTS,

surface treatment and notch radius. The results, illustrated in the appendix,

recommended to optimize the hinge taking into account these sensitivity factors.

99

Appendix

APPENDIX

400,000 373,000

350,000

300,000

Life (Cycles)

250,000

189,000

200,000

150,000

99,600

100,000

50,000

0

5 5.25 5.5

Thickness (mm)

300,000

270,000

250,000

Life (Cycles)

200,000

150,000

99,600

100,000

50,000

0

363 441

UTS (MPa)

Figure A-2 Effect of Steels UTS on the Fatigue Life of the Hinge

100

Appendix

120,000

Life (Cycles)

100,000

80,000

60,000

40,000

20,000

0

g

e

ng

d

g

le

in

on

ne

in

in

di

l

riz

at

-ro

at

N

ee

i

itr

Pl

bu

Pl

d

-p

N

ol

ar

l

ot

ke

m

C

C

Sh

ic

hr

N

C Surface Treatment

Figure A-3 Effect of Surface Treatment on the Fatigue Life of the Hinge

Notch Radius

1.50 mm

2.00 mm

2.25 mm

101

Appendix

200,000 189,000

176,000

180,000 163,000

160,000 146,000

Life (Cycles)

140,000

120,000 99,600

100,000

80,000

60,000

40,000

20,000

0

0 1 1.5 2 2.25

Notch Radius (mm)

Figure A-5 Effect of Notch Radius on the Fatigue Life of the Hinge

102

References

REFERENCES

[2] SAE International, SAE Fatigue Design Handbook, Third Edition, AE-22,

Prepared under the auspices of the SAE Fatigue Design and Evaluation

Committee, 1997.

[3] Miller, K. J., Metal Fatigue- Past, Current and Future, Proceedings of

[4] Ellyin, Fernand, Fatigue Damage, Crack Growth and Life Prediction. New

[5] Raske, D. T., and Morrow, J., Mechanics of Materials in Low Cycle

Materials and Technology, ASME, Vol. 109. October 1987. pp. 293-298.

[9] Fatemi, A., and Socie, D. F., A Critical Approach to Multi-axial Fatigue

103

References

[10] Wang, C. H., and Brown, M. W., Life Prediction Techniques for Variable

Co. 1899.

[17] Smith, K. N., Watson, P., and Topper, T. H., A Stress-Strain Function for

1969.

104

References

[23] Zienkiewicz, O. C., The Finite Element Method. McGraw-Hill Book Co.

1977.

[24] Kwon, Y. W. and Bang, H., The Finite Element Method Using MATLAB.

Algorithms, Int. Jnl. Of Fatigue, Vol. 4, No. 1, January 1982, pp. 31-40.

Keyhole Test Specimen, SAE Technical Paper, No. 750041. SAE. 1975.

105

References

[32] Tucker, L. E., Landgraf, R. W., and Brose, W. R., Proposed Technical

[33] Peterson, R.E., Stress Concentration Factors. New York: John Wiley &

Sons, 1974.

106

- Effect of Pre Straining on the Springback Behavior of the AA5754-0 AlloyUploaded byHafid Scooterist
- 05206G_TOC.pdfUploaded byHuselius
- Standart Iso 13314Uploaded byKitana Handa
- Mean Stress Effects in Stress-Life and Strain-Life Fatigue-DowlingBrasilPaperUploaded byRadwan Hazime
- Miners Rule and Cumulative Damage Models.docxUploaded byCharles Ondieki
- Optimization of the fatigue strength of materials due to shot peening : A SurveyUploaded byAman Bansal
- Din743 Intro EnglishUploaded byUber_Konstruktor
- TALAT Lecture 2703: Construction and Dynamic Testing of Aluminium NodesUploaded byCORE Materials
- tmpCF9A.tmpUploaded byFrontiers
- FatigueUploaded bySanyam Madaan
- DOEAL-SG-100_AppM_09-30-05_finalUploaded bynewavatar
- Corrosion Effects on the Ductile Fracture, Strength and Reliability of Membranes, Plates and Shells.pdfUploaded byTomás Núñez Acuña
- FHWA-ICTUploaded byJohn
- C21_2008_Myers.pdfUploaded byAhmed Taha
- Spring Design_Wayne BookUploaded bymaran.sugu
- Torsion Lab Reviewed (1) (1) (2)Uploaded byIulian Vizireanu
- CC2011_spring_design.pdfUploaded byNagarjunSH
- recos2.pdfUploaded byVincenzo Piluso
- LTRC Final Report 500 Mechanistic Evaluation of Asphalt Mixtures Containing Thiopave AdditivesUploaded bywalaywan
- 3_Zhang2016Uploaded bySPDJFHAKHDFJ
- 4[1].docxUploaded byindah0330
- Structural ResumeUploaded byEni Vino
- sTRUCTURAL DURABILITYUploaded byGuilherme Alencar
- FATIGUE TEST MBJSUploaded byKrisna Setiawan
- Design of Offshore Steel Structure General (LRFD Method) (DNV) 2004Uploaded bysimon
- 16 Rock BoltingUploaded byStajic Milan
- Prezent 02 12Uploaded byZiad Al Sarraf
- STP 91-1949Uploaded bymoonstar_dme
- Mechanical PropertiesUploaded byRahul Ramamoorthy
- chapter 1Uploaded byTejavath Kranthi Kiran99

- Lecture 9_ Twins and MartensiteUploaded byminimantis
- 00 CourseOverview Intro to OS Nonlinear v2017 Rev20170221Uploaded byRichard
- Aiv vs Protodyakonov MethodUploaded byYashwanth Kumar
- Crystal Reports_ BISAR 3 _22_1Uploaded byMohan Patil
- Leistungskatalog eUploaded byvenkatspinner
- Bond-slip Behavior of Reinforced Concrete Members(2)Uploaded bybatisxuta1
- Topology Optimization of a Car Seat FrameUploaded byem07150
- Confined Concrete Behavior Influencing FactorsUploaded byEsmerald
- Manuf SyllabusUploaded byGeorge Oliver
- Material Engineering Lab Handbook First 4 Experiments (Spring 2015)Uploaded byhadimaryam
- Spring Mass Model for Seismic AnalysisUploaded byPrakash SIngh Rawal
- Material Science Youngs Modulus CW2.docxUploaded byNasir Hadran
- Ansys Lsdyna TrainingUploaded bytt3340
- ACI-318-02Uploaded bysoroware
- Strike Slip Deformation and Their Relationship to Fractures Distributions and Basin Development-Result of Analogue SandboxUploaded byRiefqy Usman
- Bolt Design and Embedment Length CheckUploaded byHarish M P
- civil.pdfUploaded byankit
- BETONARME 1 Ders1Uploaded byArda Yıldırım
- KorisnaopterecenjaUploaded byvedran
- Problems on Simple Stress(Lec 4b)Uploaded byDaniel Naoe Festin
- Erochko_Jeffrey_A_201306_PhD_thesis.pdfUploaded byAnas Issa
- Design of Steel Beams to AISC- LRFDUploaded bysony setyawan
- uji tarikUploaded byberlian pulung
- Attachment a DR&O (2)Uploaded byNono4138
- nlefm.pdfUploaded byLuiz Cordeiro
- Civil Engineering 003613 MbpUploaded byRonal Salvatierra
- Assignment 1 MOS.docxUploaded byGaurav Raj
- pushoverUploaded byVeerendra Reddy Vintha
- EPS Foam PropertiesUploaded byShereya Singh
- strength of material Chapter 2Uploaded bynaserunn

## Much more than documents.

Discover everything Scribd has to offer, including books and audiobooks from major publishers.

Cancel anytime.